Recently, the incidence of human mycobacterial infections due to species other than M. tuberculosis has increased worldwide. Since disease control depends on appropriate antimicrobial therapy, the precise identification of these species of clinical importance has become a major public health concern. Identification of mycobacteria has been hampered because of the lack of specific, rapid, and inexpensive methods. Therefore, we aimed at designing and validating a bacterial lysate-based polymerase chain reaction identification scheme. This scheme can classify clinical isolates into: (1) the genus Mycobacterium, (2) the M. tuberculosis complex, (3) the nontuberculous mycobacteria, and (4) the species M. avium, M. intracellulare, M. abscessus, M. chelonae, M. fortuitum and M. bovis of clinical importance, and M. gordonae, the most commonly encountered nonpathogenic species in clinical laboratories. By using M. fortuitum and M. avium lysates as models, the method sensitivity was determined to be 372 pg of DNA. In a blind parallel comparison between our approach and conventional biochemical tests, both assays correctly categorized 75 patient's mycobacterial isolates. However, our approach only required 4-9 h for categorization compared with at least 15 days by conventional tests. Furthermore, our methodology could also detect M. fortuitum and M. avium from liquid cultures, after only 2 and 6 days, respectively, of incubation. Our new identification scheme is therefore sensitive, specific, rapid, and economic. Additionally, it can help to provide proper treatment to patients, to control these diseases, and to improve our knowledge of the epidemiology of mycobacteriosis, all urgently needed, particularly in developing countries.